Semidefinite Programming is currently a very exciting and active area of research. Semidefinite relaxations generally provide very tight bounds for many classes of numerically hard problems. In addition, these relaxations can be solved efficiently by interior-point methods.In this paper we study these semidefinite relaxations using the equivalent Lagrangian relaxations. In particular, the theme of the paper is to show that the Lagrangian relaxation is, in some respects, best. In all instances we consider, we show that whenever we have a tractable bound (relaxation), then the same bound can be obtained from a Lagrangian relaxation.